Metalorganic Chemical Vapor Deposition (MOCVD) is a chemical vapor deposition technique for growing crystalline layers in processes such as the production of semiconductors. The MOCVD process is implemented in a reactor chamber with specially designed flow flanges that deliver uniform reactor gas flows to the reactor chamber.
The temperature of the crystalline layers during the MOCVD process are typically measured using non-contact devices such as radiation thermometers or pyrometers. Such crystalline growth materials include silicon carbide (SiC), zinc selenide (ZnSe), and gallium nitride (GaN) based materials such as GaN and AlGaN. Certain crystalline growth materials have emission characteristics that limit the wavelength of operation for radiation thermometry. For example, gallium nitride (GaN), grown on a sapphire substrate, can have a transmittance greater than 50% for wavelengths longer than 450 nanometers (nm) at process temperatures. Thus, at wavelengths longer than 450 nm, a substantial fraction of the radiation leaving the surface of a GaN layer originates from the structure beneath the substrate that is in the line of sight of the radiation thermometer (e.g., a wafer carrier). Radiation that passes through the GaN layer is not indicative of the temperature of the GaN layer. Accordingly, radiation thermometers have been developed that detect radiation at wavelengths shorter than 450 nm (corresponding roughly to the blue, violet and ultraviolet wavelengths). See, e.g., U.S. Patent Application Publication No. 2011/0064114 to Zettler et al. (hereinafter “Zettler”), disclosing a pyrometer adapted to detect radiation in the range of 250 nm to 450 nm.
An issue with the use radiation thermometers is the detection of unwanted radiation. One source of unwanted radiation is unfiltered radiation that is detected from outside the desired band pass of detection. Zettler describes an apparatus and technique that accounts for the contribution of unfiltered radiation. Zettler points out that narrow band pass filters do not totally block infrared radiation. The unblocked infrared radiation can be problematic at the temperatures of operation (about 800° C.) because the blackbody intensity of the radiation in the infrared portion of the electromagnetic spectrum is about 9 orders of magnitude higher than in the primary band pass (i.e., the desired spectral band pass for inferring target temperature) of the narrow band pass filter. The method of Zettler involves the use of a detector that is sensitive over a broad wavelength range (from ultraviolet to the infrared) and filtering the incoming radiation with a narrow band pass filter centered near 410 nm. A longpass filter is then used to effectively block the primary band pass of the narrow band pass filter, but still allow the radiation unfiltered by the narrow band pass filter in the infrared and the near-infrared portions of the electromagnetic spectrum to pass. Zettler infers the radiation that passes through the primary band pass of the narrow band pass filter as the difference between the two measurements, i.e., between the signal attained with only the narrow band pass filter and the signal attained with both the narrow band pass filter and the longpass filter.
Another source of unwanted radiation is the contribution of “stray radiation.” Stray radiation is reflected radiation that is redirected onto the target by the enclosure or other structures therein via inter-reflection and reflected into the line-of-sight of the radiation thermometer. Consider a wafer carrier with GaN wafers that are being heated to an elevated temperature of 800° C. by, for example, a microwave heating process. The components operating at the elevated temperature, such as the wafer carrier and wafers, will emit radiation in all directions, causing radiation to inter-reflect within the chamber. Some of the inter-reflected radiation will be incident on the surface targeted by the radiation thermometer and contribute to the radiation detected by the radiation thermometer. For GaN crystalline layers at 800° C., the reflectance at 410 nm is approximately 0.2. The stray radiation contribution can significantly bias the temperature value indicated by the radiation thermometer.
Stray radiation is enough of an issue when the target is at or near the maximum temperatures within the chamber. However, when measuring radiation at or near the short wavelengths of the visible spectrum (i.e., in the blue, violet or ultraviolet wavelengths), the problem becomes exacerbated when there are other sources within the chamber that are operating at substantially higher temperatures than the target. Consider the blackbody radiation of a crystalline growth layer at 800° C. According to Planck's law, the blackbody spectral emissive power at 410 nm and 800° C. is about 2.0×10−4 watts/m2·μm. Now consider a heating source such as a resistance heating element that transfers heat to the crystalline growth layer via radiation and convection. Such a heating arrangement transfers heat in accordance with the first law of thermodynamics, which requires that the resistance heating element operate at a temperature that is significantly higher than the crystalline growth layer. An advantage of thermal radiative heating is that the radiation intensity can be tailored to have a profile across the wafer carrier that promotes uniformity of the temperature.
Consider, for example, a resistance heating element that operates at 1800° C. The blackbody spectral emissive power at 410 nm and 1800° C. is about 1.4×103 watts/m2·μm. That is an increase of about 7 orders of magnitude over the blackbody spectral emissive power at 800° C. (a typical operating temperature for crystalline growth layer during CVD operations) at the wavelength of interest (FIG. 1). Accordingly, even if only a fraction of a percent of the radiation at the 410 nm wavelength finds its way onto the detector of the radiation thermometer, the bias to the indicated temperature can be significant. Thus, the stray radiation contribution in chambers that utilize resistance heating elements can be of the same order of magnitude as the unfiltered radiation contribution identified by Zettler.
Zettler, however, is silent with respect to the contribution of stray radiation, or the effects of having radiation sources within a chamber that can effectively overwhelm the radiation that is emitted from the target. Rather, Zettler treats the target as though it is freely radiating (i.e., has no reflectance contribution). In fact, a target within a CVD chamber at that is operating at the temperatures required for crystalline growth is not freely radiating.
A chamber arrangement tailored to reduce the effects of stray radiation would be a welcomed development in the field of radiation thermometry as applied to surfaces inside high temperature process chambers.